Unicode Normalizer (NFC/NFD)

Your database search misses 'ação', a spreadsheet filter skips a row that is visibly there, or a string comparison returns false even though both sides look identical on screen. The cause is almost always the same: an accent can be stored as a single character or as the base letter followed by a combining mark. Paste the text here and see both versions side by side, with counts before and after.

There are four forms. NFC composes (é becomes one codepoint), NFD decomposes (e plus the accent, two codepoints). NFKC and NFKD do that and also apply compatibility mapping: the fi ligature splits into f and i, superscript ² becomes 2, fullwidth Ａ becomes A. That second group is one way, you cannot get the original back. The counters show UTF-16 units (what JavaScript .length returns) and codepoints, which is where the gap shows up.

For storing and comparing in a database, NFC is the usual pick on the web. NFD helps when the next step is stripping diacritics. Think twice before running NFK on technical text: m² turning into m2 and ½ turning into 1⁄2 changes the meaning of the data. This is not an accent remover, accents survive in every form. Normalization runs in your browser, nothing is uploaded.